Dell ATX PSU won’t start, but forcing it works fine?

[CapLeaker]

Got a case of something new weird at work. There is an old 2nd gen Intel i3 in an old “Pizza box” shell that has a small form factor power supply. Unfortunately I forgot to take pictures and model number off it.

Anyway, I get to work and get to look at this computer. Standby LED on the PSU is on, but the computer won't start. Hmm o.k. take the battery out of the main board try again, nothing. O.k. then… unplug the power supply from the main board and put in a full size ATX PSU (yeeeesss! Exact that blown Powerman IP that I fixed a while back), plug it in and the old pizza box works fine. O.k. Bad power supply then. So I jump the green wire to black wire on the broken PSU and it works fine as kind.
Huh? Plug it back into the old pizza box again and nothing,
I repeated that process a few times with the same outcome.
To recap: Dead PSU in the pizza box equals dead. Force the dead PSU on manually, the PSU works fine. Put a different PSU into the pizza box and the pizza box works fine.
Doesn't make any sense.

[rogfanther]

Then you opened the PSU, and found those lovely, blown caps.

[CapLeaker]

Then you opened the PSU, and found those lovely, blown caps.

Given, I didn’t check all the caps yet (due to various reasons) but the ones I did check were within tolerance.

[PeteS in CA]

Is 5VSB OK? DC voltage and ripple?

[CapLeaker]

First thing I went for. Yes. But I am thinking maybe I’ll change it out anyway, cuz sometimes that meter lies.

[lotas]

Can 5VSB handle the load without sacrificing power?

[truclacicr]

Is the PS-ON input going all the way to ground?

[CapLeaker]

Is the PS-ON input going all the way to ground?

If I put a different PSU on that computer it works fine. You would “think” the MB is fine. Have I verified exactly what happens, no. Didn’t have the time.

[stj]

the switching threshold is out of spec,
maybe cap related, you have to trace the signal

[CapLeaker]

The PSU is a DELL model L250NS-00.
I loaded the 5V STBY with 1A no problem and stable as a rock! Weird.

[truclacicr]

If I put a different PSU on that computer it works fine. You would “think” the MB is fine. Have I verified exactly what happens, no. Didn’t have the time.

That doesn't prove anything. Let's say that the PSU_ON signal sits at 1.5V instead of 0V. One PSU may see that as a logic1 while the other may see it as a logic 0.